Impact de l'intervalle d’échantillonnage sur les tests d’efficience : application au marché français des actions

Efficiency of financial markets is one of the most studied subjects in theoretical finance. Formalization of this idea is realised by the random walk model. Numerous tests have been developed to validate the hypothesis of identically and independently distributed innovations. But the results of empirical work are not unanimous; the acceptation of the random walk is not systematic and depends too much on sample features. This leads us to study the impact of the sample frequency on empirical results by distinguishing two different methodologies of tests : the first one is based on unit root (Dickey and Fuller, 1979, 1981) and the other one on variance ratio (Lo and MacKinlay, 1988). The question is the following: does rejection or acceptation of the null hypothesis of random walk depend on the sample frequency and/or on the methodology of test?